The AI Reality Check Nobody’s Talking About

Hey everyone,

I’ve been diving deep into some research that’s frankly keeping me up at night. There’s this massive disconnect between what we think AI is doing and what’s actually happening in the real world.

So here’s the deal. Stanford just dropped a study showing entry-level jobs for people in their early twenties have fallen 16% in fields that AI can touch. At the same time, MIT found that 95% of AI pilots in big companies are complete failures.

Think about that for a second. AI is already killing entry-level jobs, but most companies can’t even figure out how to use it properly.

Why Big Companies Are Fumbling This

I see this shit every day at JustPaid. Traditional companies treat AI like it’s just another software rollout. They get three people in a room, run some pilot program for six months, then wonder why nothing scales.

But AI isn’t like installing new accounting software. It requires you to completely rethink how people work. Startups get this because we’re building from scratch. We don’t have twenty years of “that’s how we’ve always done it” to fight against.

The companies that figure out scale compute, scale data, scale teams, scale adoption… they’re going to absolutely dominate. Everyone else is going to get left behind, and honestly, some of them deserve it.

The Jobs Thing Is Already Here

Customer service is getting gutted, which makes total sense. Why pay someone to follow a script when AI can do it better? The irony is that customer service has become so robotic that people sometimes think they’re talking to AI even when it’s a human.

But here’s what’s interesting about engineering roles. Yes, some basic coding jobs are disappearing. But I think people who really understand computation are about to become incredibly valuable. Not because they can write code, but because they can manage AI agents that write code.

The kids graduating now have this insane advantage if they lean into it. They’re growing up with these tools. They don’t have to unlearn decades of doing things the old way.

China’s Playing a Different Game

While we’re over here arguing about pilots and policies, China is quietly building their own AI stack. DeepSeek just started training models on Huawei chips instead of Nvidia.

This isn’t just about supply chains. This is about technological independence. China looked at our chip restrictions and said “fine, we’ll build our own.” And honestly, that’s probably the smart move.

We’re basically forcing the world to split into two AI ecosystems. I’m not sure that’s going to end well for anyone, but it’s happening whether we like it or not.

What This Actually Means

Look, the 16% drop in entry-level jobs is just the beginning. But it doesn’t have to be doom and gloom. When spreadsheets came out, people thought accountants were done for. Instead, they stopped doing basic math and started doing strategy.

The question isn’t whether AI will change how we work. It already is. The question is whether we’re going to adapt fast enough to benefit from it instead of getting steamrolled by it.

Organizations that can’t figure this out are going to die. People who resist learning these tools are going to get left behind. Countries that can’t build or access advanced AI are going to lose competitiveness.

But for the people and companies that get it right? This is the opportunity of a lifetime.

The cognitive industrial revolution is happening now. Most people are still arguing about whether it’s real.

Talk soon,

Harshith

Harshith Vaddiparthy is Head of Growth at JustPaid [YC W23] and a Council Member at Forbes Technology Council.